sieste / doit Goto Github PK

Output the normalising constant for the target density.

use nloptr and analytic expression of the gradient to speed up the optimisation

• Add argument near in doit_propose_new to specify a starting point for the maximisation of the estimation variance.

• Add argument fix to fix values for one or more parameters and only maximise the variance along the remaining ones.

Adding support for defining a fixed correlation matrix by user

Use the kernel representation to estimate the density of a linear transformation A %*% theta by a matrix A. This would include the marginals and joint marginals (#3), as well as transformations such as theta[1]+theta[2].

to illustrate 2d joint marginals

Allow multiple parameter dimensions in doit_marginal and return approximation of the joint marginal density.

doit_marginal(doit, theta_dims=c('theta1', 'theta2'), theta_eval=theta_eval)

where theta1 and theta2 are column names of theta_eval. If theta_dims is unspecified, the joint marginal over all columns of theta_eval is calculated.

• in doit_fit and doit_update check for duplicate design points
• in implement symmetric version of GGfun
• transform sum(i,j) to sum(i >= j)

• add @seealso tags
• add package meta description for ?doit
• add hyperlinks to mentioned function
• add examples